Hello Canary Team,

Thank you for all the great work on this model. I have a general question regarding its training and capabilities.

My question is about the model's out-of-the-box ability to recognize high-value proper nouns, specifically major brand names (e.g., Google, Meta, Amazon, Microsoft, etc.).

I'm trying to understand if this capability is an intentional part of the training process or more of an emergent property of the large, general dataset used. Specifically:

Was the training dataset intentionally curated or augmented to include a robust representation of major global brand names or we just rely on general dataset?

Thank you